Seven years ago, Doug Drinen wrote an article about how the "optimal draft" would look from each draft slot. Does it make sense to start your draft RB-RB-RB? What about WR-RB vs. RB-QB? My goal was to build a model to derive the optimal draft based on Footballguys' projections and the most current average draft position (ADP) data. There are some limitations to my computer processing power, however, so I need to include a bunch of restrictions. Let's get those out of the way:
- We will assume a 12-team league that has only an eight-round draft. Every team starts 1 QB, 3 RB, 3 WR, and 1 TE in the first 8 rounds (I chose that system because it almost perfectly mirrors ADP through 96 picks).
- PPR scoring system
- At every pick, you take the player available (based on ADP) projected to score the most points (based on David Dodds' projections)
Like all models, this one is imperfect. You may think that this one is more imperfect than most. That's okay, because I still think there is a lot of value in this, and we'll see why in a bit. Finding out the best way to draft in this fictional world can provide useful philosophies that you can take over to your draft.
There are 1,120 different ways to draft one quarterback, three running backs, three wide receivers, and one tight end. I went through each of those 1120 draft options for each draft slot and noted which 8 players were projected to score the most points. For space reasons, I won't list all 1,120, but I'll list the top 20 for each draft slot.
Here's how to read the table. The top combination in terms of projected fantasy points says to go running back at 1.01, then grab a RB and a WR at both the 2/3 and the 4/5 turn. At 6/7, you take your first quarterback and your final receiver, and you leave tight end for last. This is pretty intuitive, but nice to see. The formula then takes the best available player at each position for each slot, which is Peterson, White/Murray, Vereen/Brown, Kaepernick/Jennings, and finally Finley. That combo is projected to score 1,913 fantasy points. Footballguys is high on Vereen, Brown, Kaepernick, and Jennings, and Peterson/White are studs, so the next combo simply has you grabbing Gronkowski early and stealing DeAngelo Williams late (another player Footballguys thinks is undervalued.) Here are the top 20:
(Note that because the #1 and #12 teams pick at the turn, for each slot, there are 8 different identical ways of drafting. I have deleted all repeating options for all draft slots, which should make it more useful for you to view but might make some of the positional orderings misleading.)
A couple of interesting observations. Peterson doesn't dominate the #1 slot, in part because we're high on Vereen and Williams late. Of course, it's far from an optimal strategy, and not one we would recommend, but I would have expected things to look at lot worse if you took Calvin Johnson number one.
Something else that sticks out: we love Kaepernick at 72 (his ADP is 73). The top 13 combos have Kaepernick in the sixth, something to keep in mind (and perhaps cross your fingers about) if you have the #1 pick.
As you would expect, there isn't much of a difference in strategy between picks one and two. The optimal strategy is again "Best Available Running Back" followed by White/Murray, Vereen/Brown, Kaepernick/Jennings, and Finley. The next best combo simply swaps Murray and Finley for Gronkowski and Williams.
We see Romo get drafted in some of these combinations, as he shows up in our first draft that starts off with running backs and wide receivers in the first six rounds. If you are high on Finley and hate taking a quarterback early, this is a viable strategy, but it's not optimal. The main reason? That combo is identical to the number one combo, except you draft Jennings in the fifth instead of the sixth, and take Romo in the sixth instead of Kaepernick in the fifth. Obviously the calculus changes if you prefer Romo to Kaepernick, but this again highlights how we view the 49ers quarterback as a great value at the end of the sixth.
We start seeing Larry Fitzgerald show up, as the latest you can get him according to our rules is #22. You've now noticed the trend at the top of the draft: top remaining RB, top remaining WR, then Murray, Vereen, Brown, Kaepernick, Jennings, and Finley. If you are a bit riskier, take Gronkowski in the third and Williams in the eighth.
Cam Newton sneaks into some of the top picks, as he's not available at the end of round four to the top two picks (Newton has an ADP of QB46 in PPR leagues). We're pretty high on Vereen and Kaepernick, so there's no way to make up the value at running back if you draft Newton in the 4th, but the Carolina quarterback does appear in the fifth-best strategy.
Now we see things begin to shift: we see a 12-point dropoff between Charles (who has an ADP of #3 but is our #2 PPR back) and Arian Foster. As a result, Calvin Johnson/Reggie Bush begins to become a viable strategy since "Peterson/Charles" followed by "Fitzgerald/White" is off the table. Another reason that the model like Megatron here is that Footballguys likes Reggie Bush a lot, and with an ADP of 21, he isn't an option in the second round for the first three owners. The third-through-eighth round picks remain the same.
The other notable difference is that Johnson takes all 20 slots here. Again, this is a model, and all models are imperfect. But I think this shows us how big the dropoff is at running back after the first three or four are gone, and the projected value of Megatron. If you want to take a running back at #5, the optimal strategy is only the 26th best option for the 5th pick (for those curious, the combination is Ray Rice, Chris Johnson, DeMarco Murray, Cam Newton, Antonio Brown, Greg Jennings, Mike Williams, and Jermichael Finley, for a projected value of 1897.1 points). One reason for that is because Chris Johnson is such a good value in round 2 and Vereen such a good value in round 4, that the model thinks it's a waste of resources to go with a running back at the fifth spot. Note that in this system, the 5th pick is even more valuable than the first, because of the ability to get a great running back in the second round. We project Peterson (298.4) and White (259.5) to score fewer points than Calvin Johnson (332.3) and Chris Johnson (232.3).
But what happens when you can't take Megatron?
This starts to look like the combos for the first three picks: best available RB, best available WR, Murray, Vereen, Brown, Kaepernick, WR, and then Finley. The only difference here is that Greg Jennings is projected off the board at 77, so you need to have an earlier first round pick to get him in the 7th (but the dropoff to Williams is just one point). Going RB-WR is the clear strategy at six: the top strategy when you go RB-RB with this slot is actually going RB-RB-RB.
Ray Rice has an ADP of 7 (C.J. Spiller has an ADP of 6, but we have Rice projected a half-point higher), so you would think many of these combos will be identical. And, of course, the strategy doesn't really change if you choose Spiller over Rice at six, or if you're left having to take Spiller at seven. But the point of running simulations is because some results aren't intuitive. As it turns out, there is a big difference here between 6 and 7: by picking late in the third round, Murray is now off the board.
Murray is our 15h-rated back, and there's a 7-point drop from him to McFadden. That's why we recommend (in this type of league) grabbing Gronkowski in the third. That might sound like a risky pick to you, but he's still extremely valuable even if he misses three games (obviously if he starts the season on the PUP list -- and therefore misses the first six games -- that changes the math). What do you do if Murray is gone, you don't like the idea of taking the Patriots tight end, and you've been burned by McFadden too many times? Vincent Jackson began to show up in the third round for the 5th owner, but he becomes a real option for the risk-averse fantasy players here. Since we like Vereen and Williams (and you already have an RB1), and are high on Kaepernick and Finley, going wide receiver-wide receiver in the second and third rounds is a sound strategy if you can't stomach drafting a severe injury risk in the third round.
We see some of the same issues here -- Gronkowski vs. McFadden vs. Jackson in the third round -- but RB-RB is now a vaiable strategy since Alfred Morris (ADP of 17) is still available. The top four combinations all feature Dwayne Bowe (ADP of 41) in the fourth, making the first time the Chiefs wide receiver has appeared on any of these lists.
McCoy is the clear pick at #8, and I think many view him as a tier one running back. We also see Kaepernick in the 6th as a clearer strategy here for an interesting reason. One pick ago, the 7th owner had Newton in the 4th as a viable option; but here, the owner of the 8th pick can draft Dwayne Bowe, and he is a preferable pick to Newton. That forces this model's hand at quarterback, so we see Kaepernick re-emerge.
Not much changes: we have Richardson as the last Tier 1 running back (there's a 24-point drop from him to Forte), and we project Richardson as 3.5 points behind McCoy. Jason Witten wasn't an option before, but he's on the table now. The decision the 8th pick owner has to make is Witten or Bowe in the fourth versus Mike Williams or Jermichael Finley later on (or Witten/Jackson compared to Gronkowski/Bowe).
But with the top 8 running backs and Megatron gone, what do you do if you've drafting ninth?
Optimal Draft says take Jimmy Graham. The Saints tight end is the clear best player at his position, and he shows up in six of the top nine slots. Brandon Marshall is finally available in the second round, one benefit to picking this late in the first round. Make no mistake, this model says the #10 pick is at a disadvantage -- the top projection is still 5.3 points behind the top option for the #9 pick. But once the top running backs are gone, Graham seems like a solid option, especially since we're high on McFadden, Vereen, and Williams. Taking Forte is a reach -- he only shows up in three of the top twenty options, and none of the top ten --so the real debate is Graham vs. A.J. Green.
Part of that debate is Graham/Bowe at 10/39 vs. Green/Witten at those same slots. That will come down to personal preference. Alternatively, if you still want to draft Bowe in the fourth and love Green, the other option is to draft Finley at tight end instead of using that late round pick on Mike Williams. One thing that's not up for debate: McFadden is the obvious pick in round three, with Gronkowski the only other option.
Since both Graham and Green are available at #11, you would think the strategy issues here would be identical. As it turns out, the model presents an interesting question.
McFadden has an ADP of 34, so this model assumes we can't take him. That drastically lowers the projections, as he was a slam dunk pick, and now we're forced to choose between Gronkowski and Jackson. And if we take Graham, Gronkowski is off the board -- that's the reason Graham no longer becomes the top pick. Without the ability to take McFadden (because of ADP) or Gronkowski (because the TE slot is filled), your hand is forced into Vincent Jackson in the third round if you take Graham. And here's another pinball effect: Mike Williams has an ADP of 82, and we've drafted him frequently in the 7th round. But we can't do that here (the 7th round pick is number 83). That means if you don't take A.J. Green in the first round (i.e., if you pick Graham), your options become even more limited at wide receiver.
In other words, Graham may be a great pick, but taking him in the first eliminates a lot of flexibility in a league that starts only one tight end and has no flex. As a result, we see Green and Forte as the top pick in our six best strategies. There is one benefit you get to picking 11th in this model: Matthew Stafford is finally available as a 6th round pick, and he's an oustanding value there (we have him as 6.5 points ahead of Kaepernick). Since his ADP is 62, you couldn't get him this late with the 10th slot.
Alfred Morris begins to re-emerge, too. That's because without McFadden available as a safety net (clearly our model has a sense of humor) in the third round, you have to take a running back in the first two rounds. So it's either Green/Morris or Forte/Marshall or Graham/Morris. I already discussed the problems with taking Graham at this slot: either of the first two options are optimal, but there's a 7.1-point dropoff from the 10th slot to the 11th. But don't think you've solved this model yet: the 12th pick has a surprise in store for you.
Because Forte is still going to be available at 13, Graham now becomes the pick. If you've got the 10th pick or the 12th pick, taking Graham first makes sense. But if you've got the 11th pick, you can't get McFadden in the third round and you can't get Forte in the second round: as a result, Graham is a luxury you can't afford in the first round.
Taking a step back, we see that the top nine picks should be Megatron and the Tier 1 running backs. Our next three players are Graham, Green, and Forte, but Marshawn Lynch sneaks into the top twelve in ADP, which allows Forte (whom we like more in PPR leagues than Lynch) to be available at 2.01. What do you do if you pick 12th, the top nine players are off the board, and then Graham and Green are drafted? That's the worst case scenario, but you can build your team around a couple of Bears, Forte and Marshall.
More from Chase Stuart:
Defensive Team by Committee - August 8
Running Back Production by Quarter (2014) - July 29
Running Back Workload Part II - July 18
Running Back Workload - July 11
Running Back Fantasy Production in Wins and Losses - July 7
Quarterback By Committee 2014 - June 19
Rearview QB - June 5
A Starting Point for 2014 Running Back Projections - May 27
How to Project Receiving Yards In 2014 - May 14
Cross-Team Running Back Handcuffs - August 28